1. Field of the Invention
The present invention relates to surrounding image generating apparatuses and methods of adjusting metering for image pickup devices. Particularly, the present invention is preferably applied to a surrounding image generating apparatus that generates a surrounding image of a plurality of areas located around a vehicle, viewed from a virtual viewpoint located above the vehicle, and to a method of adjusting metering for multiple image pickup devices that are mounted along the perimeter of a vehicle to take images of the plurality of areas located around the vehicle.
2. Description of the Related Art
Hitherto, technologies are proposed in which multiple cameras are mounted along the perimeters of vehicles and images around the vehicles, taken by the multiple cameras, are subjected to viewpoint change to generate and display surrounding images viewed from virtual viewpoints above the vehicles. Drivers can confirm the surrounding images displayed in display devices of the vehicles to see the positional relationship between the vehicles and surrounding obstacles in order to drive the vehicles so as to prevent, for example, collision of the vehicles with the obstacles.
Generally, in systems adopting the above technologies, front cameras mounted on the fronts of the vehicles take images of “shooting areas” ahead of the vehicles. Left-side cameras mounted on the left sides of the vehicles take images of shooting areas on the left sides of the vehicles. Right-side cameras mounted on the right sides of the vehicles take images of shooting areas on the right sides of the vehicles. Rear cameras mounted on the backs of the vehicles take images of shooting areas behind the vehicles. Surrounding images around the vehicles are generated on the basis of the images ahead of the vehicles, on the left sides of the vehicles, on the right sides of the vehicles, and behind the vehicles taken by the corresponding cameras.
Since the multiple cameras are mounted at different positions along the perimeters of the vehicles in such systems, the brightness of subjects can vary depending on the shooting areas of the multiple cameras. For example, if the right side of a vehicle is irradiated with sunlight, the shooting area on the right side of the vehicle is in the sun and becomes bright while the shooting area on the left side of the vehicle is in the shade and becomes dark. While a vehicle moves backward for parking in a garage, the shooting area behind the vehicle is in the garage and becomes dark while the other shooting areas are outside the garage and become bright. In such cases, the images taken by the multiple cameras differ in luminance.
When multiple images differing in the luminance are combined to generate a surrounding image showing the entire perimeter of the vehicle, there is a problem in that breaks caused by the difference in the luminance can be seen between the images in different directions to provide an image with discomfort and poor visibility. It is difficult to combine the images taken by the multiple cameras in such a manner that any break caused by the difference in the luminance cannot be shown under various conditions differing in the position of the Sun, the weather, and the surrounding environment.
For example, Japanese Patent No. 3286306 discloses a technology of calculating a color of the pixel corresponding to a point P in the three-dimensional space, viewed from a virtual viewpoint above a vehicle, by using the colors of the point P in images that are taken by multiple cameras and that are associated with the point P in order to easily combine the images from all around the vehicle into one surrounding image so as to be close to the reality.
However, in the technology disclosed in Japanese Patent No. 3286306, there is a problem in that the color of the pixel corresponding to the point P cannot be appropriately adjusted to bring discomfort if the point P viewed from the virtual viewpoint is not associated with the images taken by the multiple cameras. In addition, there is also a problem in that it takes a significant time to perform the calculation because the color of each pixel must be calculated.